Measurement of complex signals in a newly constructed brain phantom

Abstract

This thesis presents a study of the measurements of a brain phantom designed for electroencephalography (EEG) research. The brain phantom mimics the electrical and structural properties of the human head, enabling precise control over experimental conditions. The core of this study is to provide a comprehensive set of EEG measurements where the underlying source signal distribution is known, which obviously is not the case in human measurements. To assess inter-trial and inter-individual variability, all measurements were repeated on the same day, on two different days (different cap montage) and with three different gel filling which determine the internal conductivities ("inter-individual" variability).

Initial challenges, including incomplete gel filling and unstable connections, are addressed through the design of an advanced refilling system and enhanced connectors. The study also involves the construction of complex source brain signals, which were systematically tested to ensure their richness in information and similarity to real-world neural dynamics. Using these optimized signals and an improved phantom filling, EEG data were collected, analyzed, and validated.

The research emphasizes the importance of balancing biological plausibility with experimental simplicity in signal design, incorporating features like smoothly varying frequencies and noise to mimic natural brain activities. In addition, simple decomposition algorithm, such as Independent Component Analysis (ICA), was conducted as an indicator to estimate the complexity of source signals.